[RC5] Global Stats

John Caldwell johnkc at apple.com
Fri Dec 31 11:47:55 EST 1999


Perhaps the answer is to initially consider "keys" as the unit of work 
instead of "blocks of keys."  In this context, a "key" is a thing to be 
checked, an encryption key or an OGR leaf node.

If the stats are based on 2^28 things checked, then the clients would 
report back that X * 2^28 things had been checked while completing a 
block or stub.  For encryption contests, X would generally be some power 
of 2.  For projects like OGR, X would vary depending on how many nodes 
were really checked while completing a given ruler stub.  This also 
assumes that the stats can learn to deal with fractions of 2^28 and 
breakage.

-jkc

On 12/16/99 11:09 "Jim C. Nasby" <jim at nasby.net> wrote:

>Ok, now the real question is: how do we compile meaningful
>cross-platform stats? Oh, one other caveat: we want the clients to
>eventually be able to automagically figure out what project the machine
>that they're running on would be best suited for, and the weighting
>factors used in this calculation will most probably need to be worked
>into the stats as well (it's only fair that if a client decides it's
>best to do OGR, as an example, that the blocks done by that client get
>due credit on stats).
>
>Moo!
>dB!

--
To unsubscribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
rc5-digest subscribers replace rc5 with rc5-digest



More information about the rc5 mailing list